Psych 012 Three Claims and Four Validities

Research claim

An argument based on data (measured variables)

Frequency claims

Rate or degree of a single variable

Association claims

Two variables are related/correlated

Causal claims

Claims that level of one variable is responsible for level of the other

Strength of association

Degree of predictive power

Construct validity

How well did the researcher operationalize each variable?

External validity

Would the results be the same in another population (culture, age, etc.)?

Internal validity

What if infants step off the cliff because they know they'll be caught?

Statistical validity

How well do the data support the claim that infants learned to avoid?

Example of frequency claim

4 out of 5 people think re-gifting is acceptable

Example of association claim

Infants who have been walking for longer make better motor decisions

Example of causal claim

Crawling experience causes infants to learn what actions are possible

Bad science reporting

Research: 'People who live longer also eat more nuts' - Headline: 'Eating nuts might add years to your life!'

Validity

The appropriateness (reasonable, accurate, justifiable) of a conclusion or decision.

Confounding variable

No other variable can explain the changes in the resulting variable

Direction of association

One level of a variable is likely to be associated with a level of another variable

Example of a frequency claim

32% of Americans are physically inactive and 28% of Americans are obese.

Causation

Changes, increases, reduces, prevents, promotes, affects

Association

Correlates with, is linked with, may predict, are more likely to, is tied to

Operationalized

How well a conceptual variable (construct) was operationalized.

Measuring Narcissism

Need to use a validated scale (someone has already done this for us!).

Measuring BMI

Need an accurate scale to measure weight and stadiometer to measure height; otherwise BMI won't be measured correctly.

Teacher evaluations

Fill out evals in the first week of class vs. end of class vs. after you get your grade.

Manipulated variable

If you wanted to test people's risk-taking preferences by walking over narrow bridges, varying the width of the bridge can be a manipulation of risk.

Risk

Really narrow bridges more risk of falling; risk is still a construct even though it's manipulated.

Operationalize risk

What if some people were able to cross the narrowest bridge 100% of the time? Tight-rope walkers? Wouldn't be a good way to operationalize risk.

Frequency claim

How well was the variable measured/operationalized?

Association claim

How well were both variables measured/operationalized?

Causal claim

How well were variables measured AND manipulated/operationalized?

Construct validity vs. External validity

Construct validity is about operationalizing/measuring variables; external validity is about generalization to the population.

External Validity Example

Frequency claim: '44% of Americans struggle to be happy'; were 100 college students polled? Or were participants chosen to represent different age groups and demographics?

External validity for frequency claim

Representative population across ages, levels of urbanization, SES levels etc.

External validity for association claim

Need a representative sample of children.

Causal claims and external validity

External validity gets trickier with experiments; is the sample representative? How much is the experimental situation similar to the real world phenomenon?

Confound

Something that differs between experimental groups other than the independent variable.

Internal validity application

Only applies to causal claims.

Covariance

Have an association (positive or negative)

Temporal precedence

Causal variable must occur before resulting variable

True experiment

Experimentally-manipulated conditions provide an independent variable (IV)

Random assignment

Participants randomly assigned to conditions

Dependent variable (DV)

Measure a dependent variable in an experiment

Effect size

Strength of the measured phenomenon (e.g., how strong is the correlation, how much the groups differ)

Statistical significance

How likely are the observed results due to chance?

Type 1 error

Concluding that a relationship exists when one actually doesn't (false positive)

Type 2 error

Concluding that no relationship exists when one actually does (false negative)

Validity trade-offs

The balance between internal and external validity in research

Operationalization

How researchers define and measure their variables

Causal variable

The variable that is manipulated to observe its effect on the dependent variable

Independent variable (IV)

The variable that is manipulated in an experiment

Research question example 1

Do kids learn better from educational videos when parents actively watch with them?

Research question example 2

Does doing kind acts for others make people happier?

Research question example 3

Is it easier to memorize something in silence?

Research question example 4

Does a drug improve health?